#clojure – 2009-11-05

2009-11-05

00:22	adityo	good morning folks!
01:12	hiredman	clojurebot: ping?
01:12	clojurebot	PONG!
04:00	piccolino	I am trying to use a binding as a dynamically scoped variable, but I'm getting a compile error when I reference that variable in another lexical scope (another function that gets called by the binding's form). Is there something I can do about that?
04:00	liwp	piccolino: are you talking about the (binding ..) form?
04:00	piccolino	Yeah
04:01	arbscht	piccolino: paste your code somewhere
04:01	liwp	~paste
04:01	clojurebot	lisppaste8, url
04:01	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
04:01	liwp	piccolino: and what's the error that you're getting?
04:03	tomoj	functions don't know about variables that are dynamically bound for them
04:03	tomoj	unless you tell them
04:03	piccolino	How do I tell them?
04:03	tomoj	you'd have to declare everything that might be bound
04:04	tomoj	there was a discussion about this on the compojure mailing list I think
04:04	tomoj	someone wanted to do like rails where you can just refer to instance variables in the view whether they exist or not
04:04	tomoj	and then dynamically bind them in the controller
04:04	tomoj	but the view function won't compile if the variable doesn't exist
04:05	piccolino	Ah, OK.
04:05	piccolino	That actually fixed it, thanks guys.
04:05	tomoj	hmm
04:06	piccolino	So binding doesn't do destructuring?
04:06	tomoj	if you're dynamically binding a var in another namespace, do you have to declare it fully-qualified in the other namespace that uses it?
04:08	liwp	tomoj: I wouldn't think so, if you are (use...)ng the other namespace
04:17	tomoj	well my experiments just confused me even more :(
04:32	AWizzArd	I want to typehint that my function returns an array of (primitive) bytes. Is #^bytes the right tag?
04:35	tomoj	I think it's [B
04:36	AWizzArd	funny ;)
04:36	tomoj	"[" means array, "B" means bytes
04:36	tomoj	but that's the jvm, I haven't done much typehinting
04:38	Chousuke	#^bytes should work
04:38	tomoj	how do you typehint an array of objects?
04:39	tomoj	#^"[Lthe.class.name" ?
04:42	cgrand	tomoj: #^"[Lthe.class.name;" (adde semi-colon)
04:42	cgrand	added
04:43	carpdiem	hey guys, i've been learning clojure over the past few days (coming from a python background), and i'm having a devil of a time understanding the clojure list comprehension style
04:43	carpdiem	i was wondering if anyone could help me understand it
04:43	carpdiem	the (for ...) format is just not clicking with my brain
04:44	cgrand	(for [x coll] x) is equivalent to (seq coll)
04:44	cgrand	(for [x coll] (f x)) is equivalent to (map f coll)
04:45	carpdiem	@cgrand, let me ask a simple and specific question: how would you make a list of, say, 1/n for 0<=n<100?
04:45	cgrand	(for [x coll1 y coll2] [x y]) returns the cardinal product of coll1 and coll2 (coll1 being the "outer loop")
04:46	cgrand	,(for [n (range 10] (/ 1 n))
04:46	clojurebot	Unmatched delimiter: ]
04:46	cgrand	,(for [n (range 10)] (/ 1 n))
04:46	clojurebot	java.lang.ArithmeticException: Divide by zero
04:46	cgrand	,(for [n (range 1 10)] (/ 1 n))
04:46	clojurebot	(1 1/2 1/3 1/4 1/5 1/6 1/7 1/8 1/9)
04:47	tomoj	(for [x aseq] (foo x)) is like [foo(x) for x in aseq], I think
04:47	tomoj	though I'm a python noob
04:47	yason	Apparently hinting argument types only ensures that they're java.lang.Number? Not specifically #^Integer, #^Double, etc?
04:47	carpdiem	@cgrand - thanks
04:48	carpdiem	@tomoj - that's actually really helpful, thanks for the translation
04:48	tomoj	I dunno if python's list comprehensions can do cartesian products
04:48	cgrand	yason: hinting is not casting, an hint doesn't cause the compiler to add checks to ensure a avlue is of the specified type
04:49	cgrand	s/avlue/value/
04:49	carpdiem	@tomoj - can you show me an example of a cartesian product with a clojure list?
04:50	AWizzArd	Chousuke: yes, seems to work
04:50	carpdiem	something simple, even with just a few terms
04:50	tomoj	,(for [x (range 3) y (range 3)] [x y])
04:50	clojurebot	([0 0] [0 1] [0 2] [1 0] [1 1] [1 2] [2 0] [2 1] [2 2])
04:51	yason	cgrand: I've understood that Clojure makes static, compile-time checks to ensure you don't, e.g. pass a string as some argument that's been hinted to be a number.
04:51	carpdiem	@tomoj - i see
04:51	tomoj	it binds x to the first of the first seq, then runs through the elements of the second, moves on to the second of x, etc
04:51	tomoj	though you can do more
04:51	tomoj	,(for [x (range 2) y (range 2) z (range 2)] [x y z])
04:51	clojurebot	([0 0 0] [0 0 1] [0 1 0] [0 1 1] [1 0 0] [1 0 1] [1 1 0] [1 1 1])
04:51	carpdiem	yeah, i'm starting to "see" it now
04:52	carpdiem	your python translation was super helpful
04:52	tomoj	also..
04:52	tomoj	,(for [[k v] {"foo" 3, "bar" 4, "baz" 5}] (str k v))
04:52	clojurebot	("foo3" "bar4" "baz5")
04:52	tomoj	destructuring :)
04:53	yason	cgrand: (defn mult [#^Integer a #^Integer b] (* a b)) happily accepts floats, integers, bignums etc. (Or is #^Integer/TYPE the right form? Same with that, too.)
04:53	carpdiem	@tomoj - i think it'll take me another day or two before something like that reads comfortably to me
04:53	tomoj	reading it does take a while I think
04:53	tomoj	the key is this
04:53	tomoj	,(seq {"foo" 3, "bar" 4, "baz" 5})
04:53	clojurebot	(["foo" 3] ["bar" 4] ["baz" 5])
04:54	cgrand	yason: when you hint something it helps the compiler in picking the right method calls when doing interop. When there's no interop hints change nothing to your code.
04:54	tomoj	I think python can do something like that as well
04:54	tomoj	hmm.. maybe not
04:55	carpdiem	@tomoj - ok, i'm a little confused at what just happened there
04:55	carpdiem	how did clojure know to make three pairs, rather than, say, two triples?
04:55	carpdiem	it doesn't care about the commas, right?
04:56	tomoj	right
04:56	tomoj	,(seq {"foo" 3 "bar" 4})
04:56	clojurebot	(["foo" 3] ["bar" 4])
04:56	yason	cgrand: oh, I see, it's only useful when calling Java code. Right.
04:56	tomoj	"{}" means a map, maps map keys to values
04:57	carpdiem	ok
04:57	carpdiem	but when i run (doc seq), it says "returns a seq on the collection..."
04:57	carpdiem	what is a 'seq' ?
04:57	tomoj	when you use a map as a seq, you get a seq of key-value pairs
04:57	carpdiem	just stripping out all the key-value pairs?
04:58	tomoj	there's a video you should probably watch
04:58	cgrand	yason: and once a specific method overlod is picked it's the JVM which enforces the type at runtime
04:58	tomoj	well, at http://clojure.blip.tv/
04:58	tomoj	start at the bottom of the playlist
04:59	cgrand	yason: mainly usefule calling java code but also when trying to use fast math or array ops
04:59	carpdiem	all right, i'll check that out
04:59	carpdiem	i've been going through the ociweb tutorial so far
04:59	yason	cgrand: okay. So, any purely native clojure function always just takes plain Objects as arguments? That means that i.e. integer/float arguments are always boxed?
04:59	cgrand	yason: right now yes
05:00	tomoj	seqs are clojure's big abstraction over all sequential data
05:00	cgrand	but starting from JRE 6r14 you can ask the JVM to try optimizing boing/unboxing away
05:00	AWizzArd	oh really?
05:00	AWizzArd	How would that work?
05:00	tomoj	vectors, lists, maps, sets, strings, .. whatever else there is
05:01	tomoj	you can access them all as seqs
05:01	carpdiem	ok
05:01	tomoj	and there is a library of functions which operate on seqs so that you can use these functions with all of these kinds of objects
05:01	tomoj	in (for [x foo] ...), the foo is a seq
05:01	yason	cgrand: thanks, I'll go explore more
05:01	carpdiem	thanks, i'll go through those videos now, and come back if i have more questions :-)
05:01	tomoj	(well, it's either a seq or something that can be turned into a seq)
05:02	carpdiem	gotcha
05:02	tomoj	I think those videos are not too out of date.. right?
05:05	tomoj	carpdiem: glad to have you join us :)
05:05	carpdiem	thanks
05:05	carpdiem	i'm enjoying clojure quite a bit so far
05:06	carpdiem	just need to finish wrapping my brain around the new concepts
05:06	tomoj	it's well worth it, I think
05:11	tomoj	I think the only thing I remember being out of date about the seq video is rest/next
05:12	tomoj	but that's not terribly important, you can just read the docs on those functions as they are now
05:12	carpdiem	kk
05:29	AWizzArd	cgrand: do you know how one can ask the JVM to optimize boxing/unboxing away?
05:29	somnium	is there a way to do (for [x xs] (somefn (for [y ys] (otherfn y)))) with one for statement?
05:30	cgrand	AWizzArd: -XX:+DoEscapeAnalysis
05:34	tomoj	somnium: sure
05:34	tomoj	well
05:35	tomoj	(repeat (count xs) (somefn (for [y ys] (otherfn y))))
05:35	tomoj	but I guess that's not really what you meant
05:35	AWizzArd	good, thx
05:36	somnium	well, something like (for [x xs] (apply hash-map (for [[k v] ys] [k (f v)])))
05:36	tomoj	hmm
05:36	tomoj	that still doesn't make sense to me
05:36	tomoj	you're not using x
05:37	somnium	(well pretend then function at the end is closing over x)
05:37	tomoj	so like (for [x xs] (somefn (for [y ys] (otherfn x y)))) basically?
05:38	somnium	yeah that looks right
05:38	somnium	doing (for [x xs :let [z (for [y ys] ... seems a little silly
05:38	somnium	;)
05:39	cgrand	AWizzArd: I think I read this flag should be on by default in a next release
05:39	tomoj	of course you could make one of them a map to have only one 'for'
05:39	tomoj	but that's a cheap solution
05:39	somnium	yes, but for is so readable
05:40	tomoj	one time at an interview they asked me to change some code I wrote with multiple print statements to use only one print statement
05:40	tomoj	and I gave up, couldn't do it
05:40	tomoj	the answer was to have a mutable string and replace all but the last print statements with concatenations onto that mutable string
05:40	tomoj	:'(
05:41	somnium	is that a 'clojure has made my mind immutable' anecdote, or something else altogether ?
05:42	tomoj	at that point I didn't know clojure, unfortunately
05:42	tomoj	but they asked me "can you write this with only one print statement"
05:42	tomoj	similarly you asked "can you write this with only one 'for'"
05:42	tomoj	and clearly you can by rewriting it in some other way which is essentially the same, like using map
05:43	tomoj	which obviously wouldn't satisfy you, right?
05:43	tomoj	strange thing to me is that that was the answer they were actually looking for, a trivial rewrite that doesn't actually make any difference, but strictly satisfies the question
05:45	somnium	well, in some cases map is equivalent
05:45	somnium	but if there are some more lets or filters for is better
05:46	tomoj	I still have never used those extra features of for
05:46	somnium	to my eyes they can make involved sequence transformation look so simple it seems like cheating
05:47	tomoj	but I mean we can clearly rewrite (for [x xs] (somefn (for [y ys] (otherfn x y)))) as (for [x xs] (somefn (map #(otherfn x %) ys)))
05:47	tomoj	that IS cheating :)
05:47	tomoj	I mean, it's not a real answer to your question
05:47	tomoj	(btw, I don't have any real answers :()
05:47	somnium	(for [x xs :let [z (:mykey x)] :when (< z 100)] (for ...
05:48	tomoj	essentially it seems like you're doing two nested maps, and I don't see how you could rewrite that as anything except two nested masp
05:48	somnium	yeah, I briefly wondered if a :for keyword could work, but it would at most save one keystroke from (for ... (for :)
05:49	tomoj	I wonder if (for [... :let [x ..] :when x] x) is faster than (filter identity? (map .....))
05:49	somnium	interesting question
05:50	somnium	~clojurebot
05:50	clojurebot	clojurebot is like life: you make trade-offs
05:50	somnium	~clojurebot
05:50	clojurebot	clojurebot is like life: you make trade-offs
05:50	somnium	hmm, I guess it picks responses at random
05:51	AWizzArd	cgrand: yes, that makes sense. I hope Sun will continue to improve this boxing/unboxing stuff, and maybe also add tagged numbers.
05:51	somnium	wow, the source for for is pretty hairy
05:52	tomoj	seems to be about the same
05:52	tomoj	(filter identity (map even? (range 1000000))) and (for [x (range 1000000) :let [z (even? x)]] z) both take around 2s
05:54	somnium	try macroexpanding for
05:57	tomoj	heh, yeah
05:57	tomoj	clojure's safeguards for macro hygiene don't help
05:57	tomoj	I was thinking of writing a macro to wrap around macroexpand to give a semi-readable expansion
05:58	somnium	yeah, clojure.core could be dropped at least
05:59	tomoj	when debugging your own macros I suppose it might be useful to see namespaces on some symbols
06:00	tomoj	hm, actually
06:00	somnium	I haven't had the inspiration to try to write one remotely as complicated as for thus far
06:00	tomoj	it seems swank-clojure already does this
06:00	somnium	macroexpand in place is a good way to destroy the readability of source files, I found
06:01	tomoj	(for [x xs] x) at the repl and C-c C-m with point at the beginning and you get a buffer with indented code and clojure.core removed
06:01	tomoj	though all the gensyms still make it pretty unreadable
06:01	somnium	hmm, C-m seems bound to return on my setup for some reason
06:01	somnium	what's the function name?
06:01	tomoj	and the chunking
06:02	tomoj	hmm
06:03	somnium	ah I got it
06:03	tomoj	C-c RET does it too
06:03	somnium	that's more expansion that I was hoping for
06:03	tomoj	slime-macroexpand-1
06:03	tomoj	well, that's only one macroexpand, supposedly
06:04	tomoj	C-c M-m is slime-macroexpand-all for me
06:04	somnium	looking at the source for for it seems about right
06:04	tomoj	ah yes
06:04	tomoj	then the gensyms might actually make some sense I suppose
06:05	tomoj	clojure's gensym stuff really blew my mind when I first saw it
06:06	somnium	what (still) gets me is how comparatively easy it is to write a sane macro based on a function, compared to generating a new class in python or ruby
06:10	somnium	the chunk-family doesn't seem to have doc strings
06:11	tomoj	I guess they are relatively new?
06:12	somnium	only docs I found were the assembla ticket, so I guess so
06:22	adityo	how will i use merge-with to merge maps {:a 1 :b 2} {:a 3 :c 4}
06:23	hoeck	adityo: like this, for example: (merge-with + {:a 1 :b 2} {:a 3 :c 4})
06:24	adityo	yeah, got it thanks hoeck :)
06:25	hoeck	adityo: np :)
06:53	AWizzArd	,(let [x (transient [1])] (into-array Byte/TYPE (map byte (persistent! x))))
06:53	clojurebot	#<byte[] [B@f3de46>
06:53	AWizzArd	but why:
06:53	AWizzArd	,(let [x (transient [1])] (into-array Byte/TYPE (map byte (persistent! (conj x 2)))))
06:53	clojurebot	java.lang.ClassCastException: clojure.lang.PersistentVector$TransientVector cannot be cast to clojure.lang.IPersistentCollection
06:54	AWizzArd	ah okay, nevermind
07:00	ambalek	,(+ 1 2)
07:00	clojurebot	3
07:00	ambalek	clojurebot: <3
07:00	clojurebot	<3 is </3
07:00	ambalek	clojurebot: are you saying you don't love me back?
07:00	clojurebot	Huh?
07:00	ambalek	you said </3 which kind of looks like a broken heart
07:10	somnium	clojurebot: </3
07:10	clojurebot	Excuse me?
07:18	ambalek	clojurebot: don't pretend you don't understand
07:18	clojurebot	No entiendo
07:28	aatifh	I have a map like {:a nil :b 2 :c nil :d 4 :e 5} and i want to remove all the key/val whose value is nil. Anyone in just one liner elegant way?
07:30	gerry`	(filter (fn [[k v]] (not= v nil)) yourmap )
07:32	tomoj	,(into {} (filter val {:a nil :b 2 :c nil :d 4 :e 5}))
07:32	clojurebot	{:b 2, :d 4, :e 5}
07:33	tomoj	though that will remove pairs with false as a value..
07:34	tomoj	,(into {} (remove (comp nil? val) {:a nil :b 2 :c nil :d 4 :e 5 :f false}))
07:34	clojurebot	{:b 2, :d 4, :e 5, :f false}
07:35	gerry`	,(into {} (filter (fn [[k v]] (not= v nil)) {:a nil :b 2 :c nil :d 4 :e 5}))
07:35	clojurebot	{:b 2, :d 4, :e 5}
07:38	tomoj	,(into {} (filter (comp (complement nil?) val) {:a nil :b 2 :c nil :d 4 :e 5 :f false}))
07:38	clojurebot	{:b 2, :d 4, :e 5, :f false}
07:39	tomoj	hehe
07:41	djpowell	i was just looking at sinatra, the web framework that influences compojure; but it seems hard to read without all the parens :)
07:42	djpowell	a magic token stream
07:43	_ato	djpowell: huh... you're right
07:44	_ato	that's kind of scary
07:46	_ato	Compojure is lacking an awesome logo though :p
07:46	ol3	hello, i installed clojure-mode and swank-clojure
07:47	ol3	what should i do now if I want to start normal cl slime?
07:49	tomoj	ol3: https://gist.github.com/b06ace1e874298251401
07:49	tomoj	that works for me
07:49	tomoj	M-x slime-clojure or M-x swank-clojure-project to start up a slime repl
07:49	tomoj	M-x slime starts a cl slime
07:49	tomoj	or C-u M-x slime to pick which slime you want
07:50	tomoj	_ato: http://preview.compojure.org/
07:50	tomoj	maybe it's not awesome, I dunno
07:53	ol3	tomoj: thanks, that what i was looking for
08:08	_ato	tomoj: very nice!
08:08	_ato	I hadn't seen that
08:08	tomoj	it's hidden away for some reason
08:08	clojurebot	for is not a loop
08:09	tomoj	I don't really understand the logo
08:09	tomoj	is it a martini with a blue olive?
08:11	_ato	yeah looks like a green martini with a blue olive to me
08:11	tomoj	maybe if we are classy enough to be drinking martinis, we have good composure?
08:13	_ato	I think it's a reference to Frank Sinatra
08:15	tomoj	aha
08:16	tomoj	so a hidden nod to the sinatra framework as well?
08:16	_ato	yeah, that's my guess anyway
08:24	tomoj	I have yet to use compojure for anything other than tiny experiments
08:25	tomoj	it's hard!
08:25	tomoj	compojure (and I guess sinatra as well? never used it) make you actually make decisions
08:25	tomoj	instead of having them all already made for you by a danish dude
08:39	solussd	will 'memoize' work correctly with a function that alters refs?
08:40	solussd	It seems like it should not, since the input would be just the ref, not the ref's value
08:41	chouser	solussd: changing a ref is a kind of side-effect
10:33	djork	it's correct to name .clj files with_underscores.clj
10:33	djork	right?
10:34	chouser	yes
10:34	djork	namespace/with_underscores.clj becomes (ns namespace.with-underscores)
10:34	chouser	yes
10:34	djork	k
10:34	djork	thanks
10:34	chouser	unfortunately. :-)
10:34	lisppaste8	cemerick pasted "rhickey: NoSuchFieldError on keyword lookup (regression?)" at http://paste.lisp.org/display/89834
10:35	cemerick	hrm, that doesn't happen with deftype, tho.
10:35	AWizzArd	Is there an easier for doing this:
10:35	AWizzArd	,(apply hash-map (apply concat (pmap #(vector %1 %2) [:a :b :c] [1 2 3])))
10:35	clojurebot	{:a 1, :c 3, :b 2}
10:35	chouser	AWizzArd: you want pmap?
10:36	AWizzArd	yes, no mapcat please
10:36	chouser	,(into {} (pmap #(vector %1 %2) [:a :b :c] [1 2 3]))
10:36	clojurebot	{:a 1, :b 2, :c 3}
10:36	AWizzArd	good!
10:36	chouser	,(zipmap [:a :b :c] [1 2 3])
10:36	clojurebot	{:c 3, :b 2, :a 1}
10:37	chouser	But no pzipmap or zippmap, I'm afraid.
10:37	chouser	so you're still churning out vectors.
10:37	AWizzArd	well, my %2 will be (.foo %2)
10:37	Anniepoo	anybody know how to get enclojure to start it's repl with a special java cmd line?
10:38	AWizzArd	so, the pmap is actually doing something
10:41	notallama	(zipmap [:a :b :c] (pmap foo [1 2 3])) ?
10:41	chouser	pmap doesn't appear to use chunked seqs yet.
10:41	chouser	notallama: yes, good!
10:41	chouser	rhickey: do you want a chunked pmap?
10:42	AWizzArd	good idea notallama
10:43	AWizzArd	chouser: what is a chunked pmap?
10:47	chouser	AWizzArd: if the input seq were chunked, each step would do a whole chunk instead of a single item.
10:51	chouser	this would reduce the thread-interaction overhead per item and reduce the threshold for the amount of work needed to be done on each item for pmap to be worthwhile.
10:59	AWizzArd	chouser: and what are the differences between a single item vs. chunk?
11:05	AWizzArd	git question: when I want to freshly clone the NEW branch of Clojure, how would I do that?
11:09	rhickey	chouser: chunked pmap is tricky, I've tried it a couple of times without being completely happy. The fundamental problem is you can't presume a uniformity of (chunked-.non) seq type across the iteration
11:09	rhickey	cemerick: you get that with a clean build all around?
11:10	cemerick	rhickey: yes, top to bottom, although I've now discovered that it only happens with two of the ~15 defclasses we have.
11:11	rhickey	method called bounds?
11:12	rhickey	and ant clean, not just rebuild?
11:12	cemerick	well, it has a field and a method called bounds
11:12	rhickey	cemerick: that was my quesiton
11:12	cemerick	rhickey: yes, cleaned all the way down the clojure, and back up
11:12	rhickey	is that true for other fields that work?
11:12	rhickey	field + method?
11:13	cemerick	yes, on the same defclass, there's a field + method called derived that works fine
11:13	cemerick	oh, I'm seeing a pattern, one sec
11:19	AWizzArd	What is the "git clone" command to download Clojures "new" branch instead of master?
11:19	lisppaste8	cemerick annotated #89834 "untitled" at http://paste.lisp.org/display/89834#1
11:19	cemerick	rhickey: ^^
11:19	The-Kenny	AWizzArd: After the git clone, do "git checkout new".
11:19	drewr	AWizzArd: git checkout -b new origin/new
11:20	The-Kenny	Oh sorry, origin/new. My bad.
11:20	savanni	Hey, guys, I actually have a question today...
11:22	savanni	When I try loading clojure.contrib.test-is, I get back an error saying that clojure/test__init.class and clojure/test.clj cannot be found. Any idea why this would happen when I have access to other parts of clojure.contrib?
11:26	drewr	savanni: test-is was promoted from contrib to core a couple months ago
11:26	drewr	you may have conflicting versions of both
11:26	savanni	Oh. Well that makes sense.
11:27	rhickey	cemerick: fixed
11:28	rhickey	cemerick: note this is still not full non-primitive-hinted-field support
11:28	cemerick	rhickey: meaning?
11:29	rhickey	meaning the hint is ok, but does nothing, and doesn't fail when accessed via :field
11:29	cemerick	oh, ok
11:29	rhickey	as you noted yesterday, it's not creating a String field yet
11:29	cemerick	so usages in method impl bodies will need hinting
11:29	rhickey	but the lookup thunk looked for one
11:30	cemerick	ah, I was only referring to the type on the ctor
11:30	savanni	Did test-is get move into clojure.core for clojure-1.0.0?
11:31	rhickey	cemerick: I'll have to survey how extensive a change it is to support non-primitive-local hints, it is my intention to support, but in protocol-land right now
11:40	replaca_	savanni: nope, it's post-1.0.0
11:41	savanni	Thanks. I'll need to get clojure from git, then.
11:42	replaca_	savanni: or you can get the 1.0 compatible contrib
11:43	savanni	That, too. I'm more inclined to get the latest versions, though.
11:43	replaca_	(but I just stay at HEAD and everything seems fine)
11:43	savanni	If I'm getting development versions, I'll just expect things to break occasionally. :)
11:44	chouser	savanni: you may be disappointed
11:45	savanni	You mean it rarely breaks? That would be awesome.
11:45	noidi	does anyone have any idea how to install swank-clojure at the moment?
11:46	savanni	(my statement sounds really mean, and I didn't expect it to...)
11:46	noidi	I tried following these instructions http://github.com/jochu/swank-clojure, but when I run M-x slime, I get "Symbol's value as variable is void: package-activated-list"
11:47	chouser	savanni: oh, no, it's not mean to expect a development branch to break occasionally. But in practice it's pretty rare.
11:47	chouser	and usually cgrand's fault.
11:48	chouser	:-) see, now that was mean.
11:48	chouser	cgrand: just kidding!
11:48	noidi	I guess I'll try the package management thing, maybe installing clojure-mode from git doesn't work anymore
11:48	rhickey	ouch!
11:48	fogus_	zing
11:49	cemerick	rhickey: That fix is confirmed. I'm hitting a couple of other related issues, as well. Do you want msgs on the list, tickets in assembla, or should I lay them on you here?
11:49	rhickey	cemerick: start here
11:50	stuartsierra	noidi: swank-clojure is in flux at the moment, preparing for a release; try the new mailing list http://groups.google.com/group/swank-clojure/browse_frm/thread/20141a1e19c60339
11:50	lisppaste8	cemerick annotated #89834 "untitled" at http://paste.lisp.org/display/89834#2
11:50	cemerick	rhickey: ^^
11:51	chouser	cgrand: I think I've broken things about as often, and while attempting changes far less useful or significant than transient maps.
11:52	noidi	stuartsierra, cool, thanks!
11:52	stuartsierra	noidi: no problem
11:52	noidi	I guess it's a good thing that &/¤%"#¤ github ate my detailed bug report, then :P
11:54	cgrand	chouser: well it's usually what I reply when one ask me how stable the dev branch is. "pretty stable unless one of my patches got applied"
11:55	chouser	heh
11:58	stuartsierra	Anyone want multi-maps in contrib?
11:59	lisppaste8	cemerick annotated #89834 "rhickey: issue #3" at http://paste.lisp.org/display/89834#3
11:59	cemerick	That's all I have now -- thankfully, I was able to distill good examples. :-)
12:00	AWizzArd	drewr: thanks
12:06	cgrand	about messing with core data structures: stack ops on persistent vectors can be made faster by having the last item or the two last items (depending on the vector parity) stored in fields. Thus the tail array is allocated only once every two conj (or pop) and peek is direct (no array lookup).
12:09	rhickey	cemerick: first fixed, second a limitation for now (use java-compatible field names)
12:10	rhickey	cgrand: sounds interesting, what about the cost for ordinary lookup? - definitely adds logic there
12:11	rhickey	and complexity to seq et al
12:12	lisppaste8	stuartsierra pasted "Multimaps" at http://paste.lisp.org/display/89840
12:13	cgrand	rhickey: haven't benchmarked the idea on the actual persistent vector but on a simplest prototype. The lookup cost wasn't noticeable (the indirection cost through nodes and arrays dominates I guess)
12:17	cgrand	seq support: getArrayFor(i) must return a synthetic array when i is the last or the penult (when even)
12:18	cgrand	transient ops would have added complexity
12:22	rhickey	cgrand: I don;t think I'd want to take on the complexity right now, but an interesting idea
12:22	rhickey	another idea I had was to have different types for the different depths, then there would be no recursion, straight-through lookup code per depth
12:24	savanni	What is a good wai to debug a NullPointerException, when I know which function it is happening in but not which variable or line is null?
12:26	cemerick	rhickey: confirmed that fix. Thank you very much. :-)
12:26	rhickey	cemerick: sure - keep em coming, that one was just an undone todo
12:27	rhickey	cemerick: I'm really glad you are able to try it out - you'd be amazed at how different things are under the hood :)
12:28	rhickey	e.g. completely changing the compilation of fns, (:foo x) etc
12:28	djork	oops, I just wrote clojure in my objc code
12:28	djork	(== 0 section)
12:29	cemerick	rhickey: it's been wonderfully smooth overall. Q: given that non-primitive hints are ignored right now, are host invocations in method bodies reflective? I'm not getting reflection warnings, but I can't imagine how methods are being invoked on Object fields otherwise.
12:31	rhickey	cemerick: the hints propagate through as hints, just not as storage types, so, stored as Object, resolved as direct calls due to hinting, no reflection, just checkcast
12:31	cemerick	ah, OK
12:31	cemerick	I'm really starting to like the idea of a defstatic or similar for factories.
12:31	rhickey	leave out hint you should get reflection warning
12:32	rhickey	cemerick: yes, interfaces + defstatic is near ideal external interface IMO
12:32	rhickey	as much as this isn't a full interop facility, it really lends to nice, clean designs
12:33	rhickey	and protocols will totally rock
12:33	rhickey	as soon as I can figure out MI :(
12:35	cemerick	rhickey: I had to deal with this yesterday, which was fun. Amazing what can be done in "typical Java" APIs. http://lucene.apache.org/java/2_9_0/api/core/org/apache/lucene/index/IndexWriter.html#constructor_summary
12:36	rhickey	wow
12:36	cemerick	half of them are deprecated, but nevertheless....
12:38	technomancy	lucene even seemed relatively tame for a Java API to me
12:39	technomancy	after working with Hadoop... =\
12:39	cemerick	uh, yeah, hadoop is pretty mammoth in terms of API surface area
12:40	cemerick	we've long since stopped using it. Nice concept, but way too much stuff to deal with for us.
12:40	stuartsierra	That's why I wrote clojure-hadoop :)
12:40	patrkris	does anyone here have experience with using webjure? I just want to hear some opinions on it as a web programming framework? are there any better alternatives?
12:41	cemerick	just the sysadmin junk associated with it was enough to cause us to look away.
12:41	cemerick	we're liking simple JMS with couch for data endpoints at the moment
12:41	stuartsierra	with ec2 & the cloudera scripts it's not too bad
12:42	patrkris	forget my question - i really should learn to use google
12:52	jasapp	I'm having a bit of trouble with clojure.xml/parse
12:53	jasapp	I think this is valid xml
12:53	jasapp	<foo>\"bar</foo>
12:53	jasapp	but I get an exception when I try to parse it
12:54	jasapp	,(clojure.xml/parse "<foo>bar\"</foo>")
12:54	clojurebot	java.net.MalformedURLException: no protocol: <foo>bar"</foo>
12:55	chouser	jasapp: parse wants a sax InputSource, not a string.
12:55	clojurebot	a is b
12:55	jasapp	oh, sorry
12:55	jasapp	I feel silly now, I thought it took strings
12:56	The-Kenny	,(doc clojure.xml/parse)
12:56	clojurebot	"([s] [s startparse]); Parses and loads the source s, which can be a File, InputStream or String naming a URI. Returns a tree of the xml/element struct-map, which has the keys :tag, :attrs, and :content. and accessor fns tag, attrs, and content. Other parsers can be supplied by passing startparse, a fn taking a source and a ContentHandler and returning a parser"
13:01	jasapp	ahh, much better
13:25	javuchi	hello
13:25	Joreji	Hey guys, how can I keep on calling a function over and over (in a while loop) without freezing the clojure repl?
13:25	Joreji	Is there some agent/thread magick that can achieve that?
13:26	javuchi	is there going to be a stand-alone clojure ever (with no dependences for the jvm)?
13:28	patrkris	Joreji: you can send an action to an agent which itself sends an action to the same agent, whenever it wants to continue "the loop"
13:30	AWizzArd	This works fine for me:
13:30	AWizzArd	,(let [a (agent 15), p (promise)] (send a #(do (deliver p (+ % 5)) :ok)) @p)
13:30	clojurebot	20
13:30	AWizzArd	But, when I make (+ % 5) to (+ % "5") then it hangs
13:31	Joreji	patrkris: If I do (while 1 (send my-agent (update))) wouldn't that fill the queue of the agent pretty fast?
13:31	AWizzArd	Can deliver maybe made so that when an exception occurs that it will throw it, instead of hanging?
13:34	patrkris	Joreji: just a sec, i'll give you an example
13:36	chouser	javuchi: what would you use it for with no libraries? Clojure depends on a host (JVM, CLR, etc.) for all IO, user interaction, etc.
13:38	javuchi	chouser: because the jvm consumes too much memory
13:38	wwood	4
13:39	cemerick	javuchi: so you want to target C or something?
13:39	patrkris	Joreji: http://pastebin.com/m70303ad7
13:39	patrkris	Joreji: just a very simple example
13:40	patrkris	Joreji: you don't need a while there, because every time the action is called, it will dispatch itself to the agent again and keep it going
13:41	patrkris	Joreji: and the agent queue wont fill because every time the action is executed, it is taken off the queue
13:42	Joreji	patrkris: Ah, thanks alot!
13:42	patrkris	Joreji: and the Thread/sleep is just there to pretend that we're doing some work that takes a little time
13:44	abedra	should send-off ever block?
13:46	abedra	basically i want to send an agent off every n seconds
13:46	abedra	http://gist.github.com/227280
13:46	abedra	i started here
13:46	abedra	but it blocks
13:46	patrkris	abedra: note that the action is not sent before the current action is done executing
13:46	drewr	abedra: you probably want agent on line 2
13:50	abedra	patkris: so I should create agent ahead of time?
13:50	abedra	and use that for both?
13:51	patrkris	abedra: agent refers to the agent currently executing
13:51	patrkris	abedra: it's a special variable in clojure
13:52	chouser	javuchi: once we have clojure-in-clojure, ports to other host languages/platforms will be quite a bit easier. Perhaps someone will target ObjectiveC/iPhone or chickenscheme
13:53	Joreji	patrkris: http://pastebin.com/d5bf739ca
13:53	Joreji	patrkris: For some reason I'm getting a stack overflow.
13:53	Joreji	patrkris: Anything obvious which I've done wrong?
13:54	fradiavalo	is it good clojure style to have code like this? (reduce + (take 100 (repeat 5)))
13:54	chouser	fradiavalo: it's probably better just to say (* 100 5)
13:54	chouser	:-)
13:55	chouser	fradiavalo: but other than that, there's nothing wrong with your expression. What about it makes you ask?
13:56	fradiavalo	chouser: Thanks, just wondering. I am learning clojure, so quite unsure about the correct idioms.
13:56	patrkris	Joreji: i'm not exactly sure
13:57	chouser	fradiavalo: a seq-producer (like repeat) fed through a chain of filters or processors (like take, filter, remove, map, etc.) is quite common, as is collecting up that seq with a 'reduce'.
13:57	Joreji	patrkris: I also get the stack overflow if I remove the (QApplication/processEvent) call.
13:57	chouser	there's even a macro to let you write it in what may be a more comfortable order.
13:58	chouser	,(->> 5 repeat (take 100) (reduce +))
13:58	clojurebot	500
13:58	patrkris	Joreji: do you get the stackoverflow right away?
13:58	fradiavalo	chouser: I like the ordering without the macro better :)
13:59	fradiavalo	Guess I like to read right to left
13:59	Joreji	patrkris: No. After like 3-5 seconds.
13:59	chouser	fradiavalo: fair enough
14:00	fradiavalo	I know a little bit of q (http://en.wikipedia.org/wiki/Q_(programming_language_from_Kx_Systems) so the right to left style seems more natural
14:00	notallama	fradiavalo: with the -> and ->> macros, i often put each form on its own line, so you'd read it top to bottom.
14:01	patrkris	Joreji: hmm... maybe it's because QApplication/processEvents doesn't take much time, and the agents effectively does a lot of sends really fast... but I'm not sure.
14:01	patrkris	someone else may be able to answer this
14:01	stuartsierra	Joreji: does QApplication/processEvents block?
14:01	fradiavalo	notallama: Thats a good idea to help readability
14:02	Joreji	stuartsierra: I've removed the QApplication/processEvents. So, no.
14:02	stuartsierra	then you might be overflowing the agent's event queue
14:02	Joreji	Ah.
14:03	patrkris	my thought
14:03	patrkris	exactly
14:03	Joreji	Indeed. It works if I add a (Thread/sleep 500)
14:03	Joreji	Thanks guys!
14:05	stuartsierra	Joreji: looking at the definition of QApplication/processEvents, you probably don't want to call it in a busy-loop like that.
14:05	abedra	patkris: thanks, that got me further, however I want to make the agent keep firing the function
14:05	abedra	patkris: and i get an agent error after the first couple of firings
14:06	stuartsierra	Clojure agents do not directly support calling an action periodically, you need Java timers or something.
14:06	Joreji	stuartsierra: Why is that?
14:06	patrkris	abedra: what is the error?
14:06	Joreji	stuartsierra: Hm ok.
14:06	Joreji	Explains why this is not working.
14:07	stuartsierra	Joreji: the docs say you can call processEvents "periodically", but you're calling it incessantly, like a while(1) loop.
14:07	technomancy	is there a reason you can only kill all agents at once rather than killing individual ones?
14:07	hiredman	http://gist.github.com/178350 <- schedule
14:07	abedra	thanks
14:08	abedra	patkris: i just get Agent has errors
14:08	Joreji	stuartsierra: Makes sense. Guess I'll have to take a look at some java timers.
14:08	abedra	but I think the scheduler would help that out
14:09	patrkris	abedra: use (agent-errors my-agent) to see the errors
14:09	patrkris	,(doc agent-errors)
14:09	clojurebot	"([a]); Returns a sequence of the exceptions thrown during asynchronous actions of the agent."
14:09	stuartsierra	technomancy: maybe because the JVM discourages you from killing individual threads?
14:09	hiredman	the binding of agent to thread is very loose
14:09	hiredman	an agent's action just runs on whatever thread the threadpool hands it off to
14:11	hiredman	if you want to be able to cancel async actions you might look at future
14:12	technomancy	stuartsierra: didn't realize that... (.stop (Thread. foo)) is not kosher?
14:12	stuartsierra	technomancy: The JDK docs say don't do it, I think b/c it can leave locks/monitors inconsistent.
14:13	chouser	and skip 'finally' clauses
14:13	abedra	patkris: IllegalArgumentException: Key must be integer is what gets returned there
14:14	technomancy	interesting
14:14	patrkris	abedra: well I guess you're calling something inside your action that expects an integer argument :)
14:14	Chousuke	the proper way to "stop" a thread is to have it poll for the .interrupted flag
14:15	abedra	patkris: the funny thing is that it runs twice before that happens :)
14:15	Chousuke	also if you call .interrupt on a thread while it's waiting, an exception will be raised within it.
14:15	patrkris	abedra: can you give me a pastbin of your new code?
14:19	abedra	patkris: http://pastebin.com/m4785c831
14:19	abedra	patkris: that's in the middle of trying a few things out
14:20	AWizzArd	rhickey: it seems that promises can block forever when an agent tries to deliver them with something that throws an exception. For example, paste this into your repl: (let [a (agent 15), p (promise)] (send a #(do (deliver p (+ % "5")) :ok)) @p)
14:21	patrkris	abedra: ok, first of all, the actions that you send to agents must always take the agent's state as the first argument, so you should probably just have (send-off a api-key-agent) in line 8
14:22	patrkris	abedra: and in line 4, you are asking for the agent's errors, but that will fail because a is bound to the agent's state at that point, not the agent itself
14:22	javuchi	why CL's guys still advocates against clojure?
14:23	djork	because they are smug Lisp weenies
14:23	djork	and they are no fun
14:23	stuartsierra	javuchi: because it doesn't have a 50-year-old spec
14:23	djork	and nobody likes to hang out with them
14:23	patrkris	abedra: and why don't you just call retrieve-api-keys directly?
14:24	javuchi	i found this:
14:24	javuchi	http://www.3ofcoins.net/2009/01/30/common-lisp-clojure-and-seriousness/
14:24	drewr	how do you get ns bound to the namespace of the fn's var?
14:24	rhickey	AWizzArd: the exception happens before the transfer of control to deliver. What you are basically saying is, code that throws an exception before it calls deliver doesn't end up calling deliver, which, while correct, is just a program bug, not something promise/deliver can do anything about
14:24	javuchi	all what he says is that clojure is a toy because it is hosted
14:25	javuchi	no other technical reasons
14:25	djork	" a serious language to a toy language."
14:25	djork	what a dick
14:25	abedra	patkris: That's what I would like to do...
14:25	drewr	iow, (in-ns 'foo.bar) (defn baz [x y] ns)
14:25	stuartsierra	The CL guys don't like Clojure because they see it as a concession to the market dominance of Java.
14:25	djork	the toy language is the one where I can download a jar and fire up a Repl and slap a window on the screen with the stuff already in my OS?
14:25	abedra	patkris: basically I need to have something keep calling (retrieve-api-keys) in the background while the process is running
14:26	patrkris	abedra: and you want to do something with the retrieved value, i guess?
14:26	djork	the toy language is the one where an infinite lazy seq of prime numbers is a one-liner
14:26	abedra	patkris: no, (retrieve-api-keys) just updates an atom
14:26	djork	(my favorite clojure example ever, by the way)
14:26	noidi	"after a nuclear catastrophe, archæologists of future generations should be able to re-implement the language on any hardware, including the Calculor;"
14:26	noidi	:D
14:26	AWizzArd	rhickey: ok I understand
14:26	noidi	yes, all languages should be designed for the post-apocalyptic use case
14:27	javuchi	hehe
14:27	djork	Perl and Tcl are toys too? I guess his definition of a toy language is "one that is used to do vast amounts of real work."
14:27	stuartsierra	I like this comment: http://www.3ofcoins.net/2009/01/30/common-lisp-clojure-and-seriousness/#comment-31
14:27	javuchi	i don't see the point in going against the JVM right now that it is GPL
14:28	djork	Here's the deal: the JVM is specced out, and Clojure is open source. Done. It's just as apocalypse-proof as CL.
14:28	javuchi	djork: I've never seen a serius project in CL, do you know of one?
14:28	djork	nope
14:28	javuchi	in 20 years of existence?
14:28	djork	it's just so much work to use CL
14:28	djork	there's no time left to build your project
14:29	stuartsierra	javuchi: AllegroGraph (huge triple store) was originally written in CL, then ported to Java :)
14:29	javuchi	what's going to happen to CL today, when multi-threading applications are going to be a must?
14:29	djork	I feel like clojure/core is not going anywhere... is that right rhickey? It's small and concise enough.
14:30	patrkris	abedra: do you want the agent itself to hold any particular state, or do you just want it to repeatedly call (retreive-api-keys) with certain intervals?
14:30	stuartsierra	djork: if you think clojure.core is not going anywhere, you haven't been paying attention :)
14:30	abedra	patkris: i just want to repeatedly call it
14:30	djork	I guess not :)
14:30	stuartsierra	djork: reify, deftype, transients...
14:31	djork	I don't mean in terms of new features
14:31	djork	I mean things aren't going away right?
14:31	stuartsierra	djork: there will probably not be major changes to the existing core functions any time soon
14:32	djork	"At this point in Lisp’s history a standard was still 25 years in the future."
14:32	patrkris	abedra: how about something like this: http://pastebin.com/d1eef4361
14:32	djork	(the age Clojure is now)
14:33	javuchi	ok, the question now is: is somebody doing any important project based on clojure?
14:34	drewr	,(in-ns 'foo) (defn bar [] ns) (in-ns 'baz) (require 'foo) (foo/bar)
14:34	clojurebot	#<Namespace foo>
14:34	javuchi	any significant application?
14:34	abedra	patkris: exactly like that, thanks so much!
14:34	fogus_	"Important" is a loaded term, but FlightCaster uses Clojure
14:34	drewr	,(ns 'foo) (defn bar [] ns) (in-ns 'baz) (require 'foo) (foo/bar)
14:34	clojurebot	java.lang.ClassCastException: clojure.lang.PersistentList cannot be cast to clojure.lang.Symbol
14:34	patrkris	abedra: if you need to update the state of the agent, the last expression in its action should yields its new state.... i haven't taken that into account
14:34	drewr	,(ns foo) (defn bar [] ns) (in-ns 'baz) (require 'foo) (foo/bar)
14:34	clojurebot	nil
14:34	patrkris	abedra: no problem :)
14:34	drewr	that nil seems wrong to me
14:40	mikehogye	where is the best place to ask questions about Lancet?
14:47	weissj	mikehogye: here probably :) i've worked with it some - it's not quite complete, i've made a few fixes that i don't think anyone ever accepted :)
14:48	mikehogye	weissj: in the runonce method, it seems like access to the "result" atom is not properly synced
14:48	mikehogye	weissj: http://github.com/stuarthalloway/lancet/blob/master/lancet.clj , lines 68 and following
14:49	chouser	drewr: switching namespaces within a single top-level form is perilous. You'll get different results if you run each of those in the repl.
14:49	mikehogye	weissj: ThreadA could execute line 83, find that @result != sentinel
14:50	mikehogye	weissj: then ThreadB calls reset-fn
14:50	mikehogye	weissj: then ThreadA derefs result
14:52	mikehogye	weissj: causing line 85 (and ThreadA's call to runonce) to evaluate to sentinel
14:52	weissj	mikehogye: to be honest i don't have idea idea what runonce is even for. i just require lancet as ant, and start calling things like ant/unzip, ant/move, etc
14:52	drewr	chouser: for a var that points to a fn, say foo.bar/baz, is there any way within baz to refer to find foo.bar?
14:53	mikehogye	weissj: actually, the other question I wanted to ask about Lancet is: is it usable?
14:53	drewr	I'm not switching namespaces in my real code
14:53	weissj	mikehogye: all i wanted was clojure bindings into ant, which i get without doing anything special
14:54	mikehogye	ah
14:54	mikehogye	weissj: ah, I'm with you
14:54	drewr	chouser: s/to find/to/
14:55	weissj	mikehogye: it's usable for me, but i did have to add a few multimethods like the ones that are already there, for example for exec's arg tag, and echo's level tag
14:55	chouser	drewr: not sure exactly what you're trying to do. Would this work? (defn baz [] (.ns #'baz))
14:55	mikehogye	weissj: any idea whether anyone else is actively working on it?
14:56	chouser	drewr: or: (def baz (let [ns ns] (fn baz [] ns)))
14:56	weissj	mikehogye: i emailed stuart my additions and never heard anything, and i see the current git copy still doesn't have them. so i'd say no
14:56	mikehogye	weissj: bleh
14:56	mikehogye	weissj: thanks
14:56	weissj	i think he used it as an example from his book and never expected anyone to really use it
14:56	mikehogye	weissj: that's a shame
14:56	technomancy	pretty sure lancet is intended just as an instructional project, not to be used for anything serious.
14:56	technomancy	at least, not as a tool in itself.
14:56	weissj	technomancy: yeah, which is unfortunate because it's really useful
14:57	mikehogye	it's such a good idea, though (or, it seems like a good idea to me)
14:57	weissj	and it's almost complete
14:57	weissj	then again, if its used as an example in a book, how hard can it be to finish
14:57	weissj	i haven't finished it because i'm a noob and i haven't had time
14:57	technomancy	however there is an in-process project at http://github.com/technomancy/leiningen that may be more like what you're looking for
14:57	technomancy	but you didn't hear that from me.
14:57	weissj	but mikehogye you can have my changes if you want em :)
14:58	notallama	if ::foo is a child of ::bar, will multimethods always prefer ::foo? or do i have to tell it to prefer ::foo?
14:58	mikehogye	weissj: I assume you're the weissjeffm on github?
14:58	drewr	chouser: ah, thanks; ns() is what I needed
14:59	weissj	mikehogye: yep
14:59	drewr	it'd be nice to do it without the #'baz, but at least it occurs near the scope
14:59	mikehogye	technomancy: is leiningen specifically for building Clojure codebases?
14:59	drewr	I'm trying to convert "foo" to the value that #'foo points to dynamically
15:00	technomancy	mikehogye: maaaaaaybe. (actually, yes.)
15:00	djork	someone asked about important projects in Clojure
15:00	djork	flightcaster.com
15:01	djork	built a Clojure layer on top of Hadoop and Cascading
15:01	mikehogye	technomancy: okay -- I will look at leiningen, but what I'm really dying for is Ant without the suckage
15:01	mikehogye	technomancy: which is to say, Lancet
15:01	djork	they do massive distributed computations in Clojure
15:01	djork	it is pretty much their backbone
15:02	djork	http://www.infoq.com/articles/flightcaster-clojure-rails
15:02	technomancy	mikehogye: lancet doesn't even ship with a bin script; it's definitely not meant to be used like ant.
15:04	javuchi	does clojure works on mobile devices supporting java?
15:04	javuchi	for example, on my nokia 6210 mobile?
15:04	stuartsierra	javuchi: it's reported to work on Android
15:05	hiredman	javuchi: most "java" phones are j2me which is garbage :(
15:05	mikehogye	technomancy: it seems like a basic wrapper script would be fairly easy (although I've given it a grand total of about 30 seconds of thought)
15:05	chouser	It's been a long time since I've heard anyone talk about clojure on j2me -- I don't remember if people got it working or not.
15:05	javuchi	hiredman: so it doesn't work?
15:05	hiredman	chouser: it would be really hard
15:05	hiredman	I think j2me is mostly java 1.4
15:06	hiredman	I forget, but I don't think it even has Collections
15:06	chouser	oh. ouch.
15:06	ordnungswidrig	hi
15:06	hiredman	yeah
15:06	djork	oh well... j2me devices feel slow with pure Java in the first place
15:06	djork	I'd hate to see Clojure
15:07	hiredman	I think that depends on the phone
15:08	hiredman	I did a game of life for my blackberry and it ran fine :P
15:08	ordnungswidrig	java on my sony ericsson w800i was surpringsingly fast
15:08	Chousuke	hm
15:09	Chousuke	I wonder how to make ant print out the entire stacktrace
15:09	ordnungswidrig	ant?
15:09	hiredman	I think palm pre might be the only phone with real java
15:09	ordnungswidrig	ant -d?
15:09	Chousuke	I'm trying to run contrib tests with my reader and it fails and it's hiding 80 items from me and not even -v shows them
15:09	hiredman	chouser: doesn't it usually do that at the end?
15:09	Chousuke	I get a string out of bounds exception somewhere.
15:10	hiredman	I would also get an abbreviated exception inline with the test run, and then a dump of all the stacktraces at the end of the ant run
15:11	djork	I'm happy with Clojure on the desktop right now, since it's all about the Repl anyway.
15:11	clojurebot	a is b
15:12	rhickey	woah - http://code.google.com/closure/
15:12	djork	wut
15:12	djork	you know what... the word "closure" looks funny now :)
15:12	hiredman	rhickey: yeah
15:13	chouser	,(swap! (atom [] :validator #(or (empty? %) (number? (peek %)))) conj :oops)
15:13	clojurebot	java.lang.IllegalStateException: Invalid reference state
15:14	mikehogye	technomancy: do you know of reasons that bin scripts for Lancet (or a descendent thereof) would be difficult to write?
15:15	technomancy	mikehogye: no, and that's essentially what leiningen is, plus dependency management and some built-in clojure-specific tasks. it's not hard at all.
15:15	mikehogye	technomancy: cool. I will definitely take a look.
15:16	chouser	"it compiles from JavaScript to better JavaScript" hmph.
15:16	Chousuke	hmh
15:17	Chousuke	the ant page says to use fork="true" but that's already set...
15:17	technomancy	mikehogye: once I get the jar task done I will polish it a bit to make it easier to install and make an announcement, but you're welcome to give it a go now.
15:20	rhickey	chouser: hmph? that sounds awesome to me, like hotspot for js, makes it easier for us
15:22	javuchi	how much memory consumes an agent?
15:22	chouser	I'm actually quite ambivalent about compiling to JavaScript.
15:22	hiredman	chouser: really? I thought most of the java tasks in build.xml did not have fork="true"
15:23	hiredman	er
15:23	hiredman	Chousuke:
15:23	chouser	:-)
15:23	rhickey	javascript is the new bytecode
15:23	clojurebot	new Class(x) is (Class. x)
15:23	fogus_	chouser: I've spent a lot of time in GWT, so I have similar conflicting feelings.
15:24	rhickey	unlike gwt, this seems to be the stuff they actually use
15:24	chouser	rhickey: I'll buy js is bytecode when I see debuggers that tie you back to the original source.
15:25	rhickey	good point
15:25	hiredman	hmmm
15:25	chouser	the wave team apparently uses and loves gwt
15:25	rhickey	chouser: yeah, the customized new version they had made for them addressing all shortcomings of the stuff everyone else was using
15:26	chouser	javuchi: agent uses very little memory -- it's not a whole thread or anything
15:26	chouser	rhickey: oh. :-(
15:27	javuchi	is it lightweighter tha erlang processes?
15:27	javuchi	thay consume about 230 bytes
15:27	chouser	oh, I missed that Closure is from google.
15:28	fogus_	GWT has it's nice points, when viewed through the right lens
15:28	chouser	hm, the Closure inspector sounds interesting -- that might be the missing piece for debugging. We'll just need one for clojurescript
15:29	chouser	fogus_: I can see some of the benefits, but in general I'd rather write javascript than java. :-P
15:31	fogus_	chouser: I agree with you. There is a huge impedance mismatch using GWT and that's tough to swallow.
15:32	javuchi	chouser: how can i know the size of an agent?
15:32	G0SUB	a lot of people complain about the debugging facilities of Clojure + SLIME being not so awesome. what is needed to fix the situation?
15:32	hiredman	more printlns
15:32	javuchi	G0SUB: i find it nice in the last versions
15:33	froog	http://georgejahad.com/clojure/cljdb.html
15:33	G0SUB	hiredman: better sldb integration will be great :)
15:33	hiredman	G0SUB: shrug
15:35	chouser	javuchi: http://www.javaworld.com/javaworld/javaqa/2003-12/02-qa-1226-sizeof.html :-P
15:35	G0SUB	rhickey: where can we get the video of your awesome talk about time at the jvm lang summit?
15:37	javuchi	chouser: so basically i need to create a million agents and measure memory consumption?
15:37	fogus_	Hrmmm. Reading about Google-Closure is making bitter.
15:38	chouser	agent is _meta + validator + watches + state + q + errors
15:38	djork	you mean (+ _meta validator watches state q errors) :)
15:39	chouser	each of those one a 32-bit JVM is probably 4 bytes, plus 8 for Object, I'd guess 32 bytes per agent
15:39	chouser	djork: er, right. of course.
15:39	RomanRoe	is anybody able to use build.clojure.org as mvn repo? It stopped working for me :-/
15:40	djork	my count is 88 bytes on my JVM
15:40	chouser	on a 64-bit JVM it's probably closer to 64 bytes.
15:40	javuchi	so it is expected to use 64 bytes on a 64bit machine
15:40	djork	OS X Snow Leopard
15:40	chouser	djork: counted how?
15:40	javuchi	88 bytes?
15:40	javuchi	thats much less than 230 bytes for process in erlang
15:40	djork	(- (mime (agent nil)) (mime nil))
15:41	djork	err that doesn't really work right now
15:41	djork	it's my new toy memory measuring macro
15:41	chouser	ah
15:42	djork	but (mime nil) takes 24 bytes to evaluate, and (mime (agent nil)) takes 112 bytes to evaluate
15:42	chouser	well, that's about as accurate as it's worth trying to get -- something less than 100 bytes per agent.
15:42	djork	yes
15:42	javuchi	so it is reasonable to think that we could use around 60 millions of agents with 4gb of memory?
15:42	javuchi	perhaps 50 millions?
15:43	chouser	I don't know what all an erlang process has to track, but agents don't have their own stack or instruction pointer or anything like that.
15:43	chouser	An agent is just a simple one-element container with a queue of actions to perform.
15:43	tmountain	GOSUB: http://www.infoq.com/presentations/Value-Identity-State-Rich-Hickey
15:44	chouser	javuchi: as you queue up more actions, the agent will hold onto more memory until it can dequeu them.
15:44	djork	that is some zen stuff there
15:44	javuchi	so 50 millions is not very reallistic
15:45	javuchi	perhaps 5 millions would be more reallistic
15:45	danlarkin	javuchi: there's no way to answer that, it depends on what the agents are doing
15:45	djork	I would think about my architecture as soon as 5 million of anything was being considered.
15:46	javuchi	djork: think, for example, of a massive IM server
15:47	javuchi	that's when erlang is brillant
15:47	javuchi	i'm think if clojure might be better???
15:47	dysinger	re: bulid.clojure.org RomanRoe I had to change the url and hadn't got around to posting an updated message to the list
15:47	ordnungswidrig	javuchi: but 5 million on a single node / server sounds too much even for erlang
15:48	dysinger	the maven url is http://build.clojure.org/snapshots now RomanRoe
15:48	hiredman	javuchi: clojure agents are not actors
15:48	djork	a massive server probably needs to be distributed, whereas clojure's strong point is concurrency
15:48	dysinger	RomanRoe: I wont mess with it again
15:48	djork	you could load balance between clojure nodes
15:49	javuchi	anyway clojure is not designed for being distributed, is it?
15:49	hiredman	2009:Sep:22:09:02:37 rhickey : Writing a message queue? use Erlang. Writing an ordinary application? use a message queue.
15:50	RomanRoe	dysinger: thanks!
15:50	dysinger	I still like erlang too
15:50	RomanRoe	dysinger: my clojure hacking evening is safe again.... puhhhh..... ;-)
15:51	dysinger	there are things you can do with networking and clustering in erlang that are trivial compared to socket programming in java land.
15:51	djork	indeed
15:52	djork	I'd say that if you want to, just develop your server in Clojure and then try scaling it with a message queue or something like Hadoop.
15:52	djork	It would be a good experiment.
15:52	djork	I don't see it going too badly either.
15:53	dysinger	I am working on a peer-2-peer work group library in clojure that will allow clusters of repls to work together. Message queues don't solve all problems neatly.
15:53	djork	mmm distributed repls
15:54	hiredman	dysinger: sounds kinda like terracotta
15:54	dysinger	minus 300k lines of code and instrumenting and XML
15:54	hiredman	:P
15:55	dysinger	no I don't have any shared memory needs
15:55	dysinger	I just need lots of machines to work together
15:55	hiredman	http://vids.myspace.com/index.cfm?fuseaction=vids.individual&videoid=60771080 <-- Origins of APL, the hair styles alone are worth it
16:00	ordnungswidrig	hiredman: great video :-)
16:00	ordnungswidrig	event roger moore has a role in it
16:01	ordnungswidrig	s/nt /n /
16:14	chouser	they seriously didn't pause for a moment when they realized the development of their language required new keyboard mappings, fonts, and a "type ball"?
16:17	ordnungswidrig	chouser: they where from IBM, so it was no issue to build some new hardware.
16:17	ordnungswidrig	chouser: I think
16:18	froog	besides, the world only needed 5-6 computers
16:18	froog	back then
16:18	ordnungswidrig	with a most 640k memory
16:20	ordnungswidrig	the host is smoking while the guest is talking. different times then.
16:20	dublindan	how can i force symbol capture in macros?
16:20	chouser	dublindan: ~'foo
16:20	clojurebot	foo is is Short, Self Contained, Correct (Compilable), Example http://sscce.org/
16:20	dublindan	chouser: thanks!
16:25	javuchi	dysinger: do you have something done?
16:25	dysinger	it's a wip - soon
16:25	dysinger	I love it :) so simple and so cool :) can't wait to share
16:26	javuchi	would it bring the power of erlang distributed to clojure?
16:27	chouser	javuchi: http://github.com/amitrathore/swarmiji
16:29	javuchi	looks nice, chouser
16:29	javuchi	dysinger: is your library something like that?
16:29	dysinger	no - that's message queues with no job tracking
16:30	dysinger	and no peer awareness
16:30	dysinger	not to knock it - amit is smart
16:30	dysinger	it's just a different tool than I am targeting
16:31	javuchi	so what are you doing?
16:34	javuchi	dysinger?
16:37	dysinger	javuchi: above "peer-2-peer work group library in clojure that will allow clusters of repls to work together."
16:38	rhickey	dysinger: are you using something like JGroups under the hood?
16:38	dysinger	IE, not message queues but clustered interacting repls.
16:38	dysinger	rhickey: yes-ish + branch/leaf node work tracking
16:39	rhickey	cool
16:39	javuchi	dysinger: i'm trying to figure out what exactly you mean
16:39	dysinger	just give me a couple weeks more I'll get it out there O.S.
16:39	danlarkin	dysinger is spilling the secret sauce!
16:39	dysinger	:)
16:40	rhickey	dysinger: thanks for splitting off new branch on build.clojure.org - much easier to see what is what
16:41	dysinger	yeah plus it pollutes the maven repo with mixed branches plus there is a bug in hudson that once you add a branch you can't remove it.
16:41	dysinger	so it's best to just "copy" the job and modify it for the other branch
16:56	javuchi	what do you think about Shoal?
16:56	javuchi	(for clustering)
17:01	twbray	Gosh, it turns out that when you have to say -Xmx900m to get a clojure app to run, and it has tons & tons of threads, it really sucks the life out of your basic MacBook.
17:02	hiredman	which one is Xmx again?
17:02	chouser	one of the unsung benefits of single-threaded apps -- no one of them can completely saturate your dual-core machine. :-)
17:03	hiredman	~google Xmx
17:03	clojurebot	First, out of 66500 results is:
17:03	clojurebot	XMX Home Page
17:03	clojurebot	http://www.cs.brown.edu/software/xmx/
17:03	hiredman	:\|
17:03	danlarkin	it's max heap
17:04	hiredman	clojurebot: Xmx is max heap
17:04	clojurebot	Alles klar
17:13	leafw	twbray: if your mac has two cores, try -Xincgc (incremental garbage collector). I found it to make the JVM perform better when creating lots of throwaway objects like clojure does.
17:14	twbray	leafw: Cool, will give it a try. But this thing isn't really designed to run on my laptop :)
17:15	leafw	as for clogging your poor macbook, all one can do is renice the process. Other things help too to avoid swappign: remove window shading, remove window double buffering, closing power hungry apps like MSWord and Photoshop.
17:15	leafw	twbray: then ssh to a server :)
17:15	twbray	MSword!?!? Quit talking dirty, this is a respectable IRC channel
17:15	leafw	twbray: I run ubuntu here on a macbook pro 5,5
17:16	leafw	twbray: M$word is used by just about everybody I know. And I know it to be a problem, bot memory wise and because it can't be really swapped out; it keeps constant activity
17:17	twbray	leafw: You need to start hanging out with a better crowd :)
17:17	leafw	twbray: the world is larger than you think.
17:21	javuchi	would clojure work with terracotta?
17:21	chouser	javuchi: yes, more or less. There's been some work in that direction, not sure of its current state.
17:29	javuchi	this seems to be the most up to date information?? http://paul.stadig.name/2009/03/clojure-terracotta-next-steps.html
17:31	jbendig	javuchi: Here is someone who uses Clojure with Terracotta on a live system for a vet hospital. http://groups.google.com/group/clojure/browse_thread/thread/85649a134cf22655?tvc=2
17:34	Chousuke	hmmhmm.
17:39	Chousuke	hoh
17:39	Chousuke	Finally managed to compile contrib
17:40	hiredman	:D
17:40	Chousuke	this is a great step forward :P
17:42	hiredman	so even after you could read clojure, contrib still had problems?
17:43	Chousuke	reading is different from reading it correctly so that it can be compiled :P
17:44	hiredman	I mean, you can read clojure.core in and it compiles fine, yes?
17:46	Chousuke	yeah. though I haven't tried to use the compiled result
17:46	Chousuke	... much :D
17:48	savanni	I asked this before but may have missed the answer, so I'll ask again. Is there a way to enable a stack trace or something for when I get a Null Pointer Exception in my code?
17:49	Chousuke	you can do (.printStackTrace *e)
17:49	Chousuke	if in repl
17:49	Chousuke	*e means latest exception
17:51	savanni	Awesome. A little arcane, but that will help out a lot later.
17:51	Chousuke	hmm, contrib tests pass, except for one that overflows the stack for some reason
18:00	hiredman	savanni: checkout clojure.stacktrace
18:03	savanni	Is that in core?
18:03	hiredman	it's in stacktrace.clj, which is right next to core.clj
18:04	savanni	Okay... like with test-is, I'm going to need to upgrade to the clojure dev before I will be able to use it.
18:04	hiredman	are you sure? I'm not sure when it was added
18:05	savanni	Well, my 1.0.0 library claims that clojure.stacktrace is not found, so I would assume so.
18:39	cow-orker	is there a practical difference between #_"comment" and ;comment ? I know they have different "scope", but for short comments...?
18:40	hiredman	#_ eats the following form
18:41	hiredman	; drops everything until a newline
18:41	hiredman	,#_(
18:41	clojurebot	EOF while reading
18:41	hiredman	,;(
18:41	clojurebot	EOF while reading
18:41	hiredman	not a good demo, I guess
18:43	cow-orker	well.... I got that part more or less :) I asked because #_"" was used for short comments in (defmacro for ...) in core.clj. Just wondered if there was some reason for commenting that way.
18:43	Chousuke	well, you can put them in the middle of stuff
18:43	Chousuke	,[foo #_"bar:" bar]
18:43	clojurebot	java.lang.Exception: Unable to resolve symbol: foo in this context
18:43	Chousuke	oops
18:43	Chousuke	,'[foo #_"bar:" bar]
18:43	clojurebot	[foo bar]
18:44	cow-orker	:)
18:44	Chousuke	,'[foo ;bar: bar]
18:44	clojurebot	EOF while reading
18:48	cow-orker	what got me wondering was in master/src/clj/clojure/core.clj line 3019, but I guess here it's just style
18:50	hiredman	interesting
18:51	hiredman	I think a lot of that is chouser's work
18:51	hiredman	(maybe all?)
18:53	cow-orker	thanks, I'll ask him when he's online :)
19:07	Joreji	Hey guys, is there some way to ensure that an agent is running in the main thread?
19:08	rhickey	Joreji: will never be
19:09	rhickey	they run in thread pools
19:10	Joreji	Oh.
19:10	Joreji	Ok, thanks.
19:19	Joreji	How can I translate "new TimerTask() { public void run() { ... } }" into clojure?
19:20	Chousuke	can you construct TimerTasks from callables?
19:21	Joreji	In java? Yes.
19:21	Chousuke	Then you could just do (TimerTask. somefn)
19:21	Chousuke	eg. (TimerTask (fn [] (println "hello"))
19:21	hiredman	no, you can't
19:21	Joreji	But that would require me to write java code.
19:22	hiredman	Joreji: Chousuke meant "Does the TimerTask constructor take a Runnable"
19:22	hiredman	which it doesn't
19:22	Joreji	Ah, no.
19:22	Joreji	It does not take a Runnable.
19:22	hiredman	wrapping a fn in a proxied TimerTask is pretty trivial
19:23	Chousuke	I gues you'll need proxy then :)
19:23	Chousuke	+s
19:23	hiredman	I would push you in the direction of ScheduledThreadpoolExecutor
19:23	hiredman	which takes runnables directly
19:24	Joreji	hiredman: Hm. Are the java timers all running in a different thread?
19:25	onigiri	hi guys
19:26	onigiri	I'm really really new with clojure, I was playing along with the REPL now and tried some functions:
19:26	onigiri	I've got a utf8/strings bug (or at least what I think is a bug)
19:26	onigiri	(.toUpperCase "harajuku 原宿") yields "HARAJUKU 僆宯"
19:27	hiredman	Joreji: Timer uses a single thread
19:28	Joreji	Hm ok.
19:28	onigiri	I'm still trying to wrap my head around functional programming, so I thought that the only thing I was able to do was to report it ;)
19:28	hiredman	onigiri: it might be an issue with java's String
19:28	hiredman	in which case there is nothing to be done about it
19:29	Chousuke	onigiri: I don't think that's a Clojure bug :)
19:29	onigiri	hiredman: oh ok :)
19:29	Chousuke	amusing, thoug.h
19:29	onigiri	Chousuke: heh
19:30	Chousuke	I guess toUpperCase is just a dumb function that assumes its input is actually something that can sanely be uppercased :P
19:31	onigiri	and the fun thing was that I guess was twbray's fault if I'd tried it :P I remember his talk about unicode and ruby and since he's now learning Clojure, my head mixed the 2 thoughts :P
19:31	onigiri	Chousuke: yes, I guess so.
19:32	onigiri	I may peek under the clojure hood to see if there's a hint though
19:32	technomancy	onigiri: there's not much of a hood; that's a direct JVM method call. =)
19:32	twbray	onigiri: That is kind of surprising
19:32	Chousuke	Java strings aren't quite as bad as Ruby (1.8) "strings" but they're not perfect either :/
19:32	twbray	String.toUpper looks at your LOCALE and does all sorts of voodoo
19:33	cemerick	casing is very complicated
19:33	onigiri	technomancy: oooh! you're right! I was in fact reading the direct java call
19:33	onigiri	paragraph...
19:33	twbray	but it should realize that that's in a case-free language and just no-op
19:34	onigiri	twbray: yes, indeed: before asking here I tried with ruby and it worked fine (I tried just to check if I was expecting the correct behaviour)
19:34	twbray	HOld on, I keep a java sandbox project around for just this sort of thing
19:35	cemerick	onigiri: http://www.brosinski.com/stephan/2008/02/09/the-evils-of-javalangstringtouppercase/
19:35	twbray	onigiri: Remember, in my original Ruby/Unicode talk I basically said "Don't use toUpper/toLower". I still think I was right :)
19:35	cemerick	if you provide the target locale, it will likely work correctly
19:36	onigiri	twbray: hehe I know :) I still have that talk burned in my head :)
19:36	onigiri	cemerick: reading now, thanks
19:37	twbray	onigiri: When I run it in pure Java in NetBeans, it comes out OK
19:39	onigiri	twbray: hmm, if you run in your REPL is the "bug" reproduced?
19:40	twbray	user=> (println (str "In: harajuku 原宿 out: " (. "harajuku 原宿" toUpperCase)))
19:40	twbray	In: harajuku 原宿 out: HARAJUKU 原宿
19:40	cemerick	onigiri: what's your system/java locale?
19:40	twbray	Right, try echo $LOCALE at shell
19:41	onigiri	not set
19:42	onigiri	twbray: I'm on a MBP 10.5, and if I run locale, I get LANG="it_IT.UTF-8"
19:43	onigiri	but an echo $LOCALE yields a deafening void
19:43	twbray	As long as it's got UTF-8 in there you should be OK.
19:43	cemerick	onigiri: eval this in clojure: (java.util.Locale/getDefault)
19:43	onigiri	twbray: you on SLeopard?
19:44	twbray	onigiri: yes
19:44	twbray	Java 6, 64 bit
19:44	onigiri	cemerick: hehe here's the culprit
19:44	onigiri	user=> (java.util.Locale/getDefault)
19:44	onigiri	#<Locale en_US>
19:44	twbray	onigiri: Same for me, en_US
19:45	cemerick	yeah, that's mine, I don't have an issue with uppercasing those glyphs
19:45	cemerick	chars*
19:45	twbray	onigiri: What environment are you running in? xterm? IDE?
19:45	onigiri	twbray: plain old Terminal
19:45	twbray	I'm doing this in NetBeans Enclojure REPL
19:46	onigiri	a github cloned clojure just builded
19:46	onigiri	that was run with a java -jar clojure.jar
19:47	twbray	Hah! When I do it terminal, I see the bug
19:48	twbray	Mac Terminal.app I mean
19:48	onigiri	twbray: good, at least I'm not seeing things ;)
19:48	cemerick	funny. Just set your encodings in terminal's prefs, probably
19:51	twbray	onigiri: Go get enclojure, it's pretty good
19:51	twbray	Disabled everything but UTF-8 in the terminal prefs, didn't seem to make any difference.
19:52	onigiri	twbray: I read your post, I was more partial to try the vimclojure
19:52	onigiri	I don't have any java IDE or emacs installed
19:52	twbray	That might work. Still can't figure out what Terminal.app is sending Clojure though.
19:52	cemerick	twbray: glad you've come around on enclojure :-)
19:53	onigiri	twbray: for what it matters, I tried setting UTF16, but nothing changed
19:53	twbray	oh, of course, inside NetBeans, it's probably an all-UTF16 world.
19:54	cemerick	I presume those are display encodings. It'd be worthwhile to see what code point clojure reports in the terminal for those chars, maybe the input is getting munged.
19:55	twbray	cemerick: I think that has to be it.
19:56	twbray	onigiri: Anyhow, the right answer is still the same - don't use toUpper/toLower. (snickers)
19:57	twbray	Among other things, they're unbelievable performance pigs.
19:57	onigiri	twbray: hehe
19:58	cemerick	(int (first "宿")) reports 194 in my terminal, 23487 in enclojure, definitely an input issue
19:58	cemerick	and (count "宿") is 3 :-)
19:58	onigiri	at least this bug made me do a foray in the #irc and now this channel is set for autojoin ;)
19:59	Chousuke	I get 23487 in emacs
19:59	cemerick	,(.toUpperCase "宿")
19:59	clojurebot	"宿"
19:59	Chousuke	I have it configured to be UTF-8
19:59	cemerick	it's crazy that irc is so unicode-friendly these days
19:59	onigiri	so the Terminal.app is the borked one
20:00	twbray	I usually drop into an aquamacs shell when I have to do serious non-ASCII at the command-line level
20:00	Chousuke	cemerick: well, at least English-speaking people won't have to deal with encoding wars because UTF-8 is ascii-compatible :P
20:00	cemerick	Chousuke: sorry? :-)
20:01	twbray	Maybe iTerm would be better?
20:01	Chousuke	cemerick: some years ago there were debates all over the Finnish IRCSpace whether UTF-8 should be allowed or not, because the de-facto standard was ISO-8859-1 and it's UTF-8-incompatible.
20:02	Chousuke	onigiri: I think you need to specify the encoding with a jvm property somehow.
20:03	twbray	Debates about i18n stuff used to regularly provoke paralyzing stupidity up until sometime earlier this decade when everyone just rolled over and gave in to Unicode.
20:03	onigiri	Chousuke: I fear so
20:04	cemerick	Chousuke: I'm guessing it has nothing to do with java.
20:04	Chousuke	I get 194 in iTerm as well, with a plain repl.
20:07	Chousuke	If I do java -Dfile.encoding=UTF-8 -jar clojure.jar then it works. .)
20:07	cemerick	huh
20:08	Chousuke	it defaults to MacRoman on OS X I think
20:09	Chousuke	so when I print UTF-8 in Clojure in an UTF-8 terminal it'll show up even then, but only because the bitpattern happens to match what the terminal expects, not because the JVM prints it out as UTF-8 :)
20:09	onigiri	Chousuke: oh you beat me: I just found the -D flag ;) nice
20:11	onigiri	so I guess the matter was solved, thanks everyone!
20:11	Chousuke	I wonder why the JVM is MacRoman by default anyway :/
20:11	Chousuke	OS X is pretty much entirely Unicode otherwise :P
20:22	onigiri	ok, back to clojure study
20:53	krumholt	hi can i proxy an abstract class?
20:54	cemerick	yes
20:55	krumholt	so if i proxy something it will be an anonymous class extended from the abstract class?
20:56	cemerick	right
20:56	krumholt	ok thanks
21:01	djork	what's the most idiomatic way to do something n times?
21:01	djork	I've been doing (take n (repeatedly #(foo)))
21:02	djork	but it seems like the closure is a bit unnecessary
21:02	cemerick	dotimes if you're looking for side effects
21:03	mikehinchey	you can just use foo instead of #(foo) if foo is a regular function
21:04	cemerick	hah, I didn't even notice that :-P
21:04	cemerick	,(take 5 (repeatedly rand))
21:04	clojurebot	(0.2702439022277704 0.06886273089068606 0.908758027809942 0.5910756151509906 0.9058909510000488)
21:04	cemerick	I wrote that a couple of times last year :-)
21:05	djork	I am looking for dotimes, I think
21:05	cemerick	djork: note that it isn't lazy and returns nil
21:06	djork	not a problem
21:06	djork	just need to literally do something
21:06	djork	purely for side effects
21:06	icey	Anyone here spend much time with compojure? I'm having an issue that doesn't make sense to me when trying to include it in slime. I'm sure it's a noob issue.
21:06	hiredman	what is the issue?
21:07	icey	(I'm getting NoClassDefFoundError on org/apache/commons/fileupload/FileUpload whne I try to (use 'compojure))
21:07	hiredman	ah
21:07	djork	(dotimes [_ 5] (System/gc))
21:07	icey	compojure is in my classpath, I am getting the problem on 2 machines, so whatever I've done wrong I've done wrong in a consistent fashion
21:08	hiredman	just a second
21:08	cemerick	djork: invoking System/gc is generally a waste of time
21:08	hiredman	do you have all of compojures dependencies on the classpath?
21:08	icey	hmm most likely not
21:08	djork	cemerick: not in this case
21:09	djork	it's for a very particular purpose
21:09	hiredman	well, there you go
21:09	hiredman	if you run ant it will grap a zip of all the deps
21:11	icey	hiredman: i'm taking a look in the deps dir, just so I'm clear all the jars need to go in the classpath? (sorry, I have done like 0 java prior to touching clojure)
21:12	hiredman	yes
21:12	icey	hiredman: i'll give it a swing and see what happens, thanks :D
21:12	hiredman	you can leave out clojure/contrib if you already have those, but you might run into versioning issues
21:29	icey	hiredman: thanks for the tip regarding the classpath stuff on compojure; that did the trick for me :D
21:37	djork	the classpath is a harsh mistress
21:47	djork	anybody using something like this? http://gist.github.com/227662
21:47	djork	is this adequate?
21:47	djork	I use it in my clj script
21:53	_mst	the 1.6 JVM actually lets you do java -cp '/some/where/*' to pick up all .jars from that directory
21:54	_mst	but I've used that same trick in 1.5 :)
21:54	djork	oh wow that's nice
21:54	djork	I'm using 1.6
21:54	djork	oh well it's already working
21:54	djork	I can also do . src foo/*.jar to build it
21:54	djork	real globbing is nice
22:56	qed	I know this is kind of a lame question but-- I'm wondering if there are any easier ways than using compojure
22:56	qed	has anyone built an all-in-one installer or package for compojure?
23:08	hiredman	qed: running ant in the git checkout of compojure will download all the dependencies
23:30	qed	hiredman: it's not working for me
23:30	hiredman	what is not working?
23:31	qed	/Users/defn/src/compojure/build.xml:35: /Users/defn/src/compojure/deps not found.
23:31	qed	no deps dir in there
23:31	hiredman	make one?
23:31	qed	omg im a moron
23:31	hiredman	…
23:31	qed	apologies
23:32	qed	PS: I didn't need the UTF-8 to rub it in
23:39	technomancy	it would really be great if contrib's 1.0-compat branch got its own version number to distinguish it from master so we could add it to build.clojure.org
23:39	technomancy	could we bump master up to 1.1.0-alpha-SNAPSHOT to match clojure itself?